Machine Learning Algorithm
Introduction to Machine Learning and its algorithm
What is Machine learning?
Machine learning can be defined in a variety of ways. Arthur Samuel termed “machine learning” as a “field of study that gives computers the capability to learn without being explicitly programmed”. In very simple terms, machine learning can be defined as improving the learning process of a computer without any human assistance. It is the subfield of Artificial Intelligence.
Types of Machine learning algorithms
Machine learning algorithms are commonly divided into four main categories.
· Supervised Learning
· Unsupervised Learning
· Semi-Supervised Learning
· Reinforcement Learning
Supervised Learning
Supervised learning is when a model is trained with a labeled data set. A labeled set of data means some data tagged with the correct answer. After the model has been trained, it is fed new data without the target variable, and the model predicts the target labels based on its learning. There are two types of supervised learning models.
Regression
The process of anticipating results that are continuous in nature is known as regression. Continuous outputs are integers or floating-point values. For example, predict the prices of houses based on numerous parameters such as the number of bedrooms, built-up area, number of bathrooms, etc., weather forecasting, and so on.
Algorithms for Regression
· Linear Regression
· Support Vector Machine
· Decision tree
Classification
Classification is the process of predicting results that consist of discrete output like “yes” or “no”, “spam” or “non-spam” etc.
Algorithms for Classification
· Naïve Bayes
· Logistic Regression
· K Nearest Neighbour classification
Unsupervised Learning
Unsupervised learning refers to identifying patterns in a dataset that are not labeled. It enables the model to recognize patterns on its own, without the need for human intervention.
The data that is used for training is unlabelled. There are two types of unsupervised learning: “clustering” and “association.”
Clustering
Clustering is the process of putting together groups of objects that are similar in some way. The algorithm used for clustering is K means clustering.
Association
Association is the process of discovering rules that describe the data. For example, People who buy X, items are more likely to buy Y items.
Semi-supervised Learning
We can see that the data is either labeled in supervised learning or unlabelled in unsupervised learning in the first two kinds. Semi-supervised learning falls in between these two types. In semi-supervised learning, the data involved contains less labeled data and more unlabelled data. This can happen when the cost of labeling the data is too high.
Reinforcement Learning
Reinforcement learning is the process in which an agent learns to behave in an environment based upon the rewards received. These rewards can be positive or negative. For example, the positive reward for good action and a negative reward for bad action. An agent learns through experience.
Terms used in reinforcement learning
1. Agent- An entity that investigates its surroundings and acts in accordance with its findings.
2. Environment- Space in which the agent would act.
3. Action- Moves taken by the agent in the environment
4. State- Situation returned by the environment after every action taken by an agent.
5. Reward- A feedback returned by the environment to the agent.
6. Policy- An agent’s plan for taking the next action depending on the present state.
Applications of reinforcement learning include self-driving, board games (chess), etc.
Algorithm for Reinforcement learning
· Value-Based
· Policy-based
· Model-based
